Julian D. Marshall’s research while affiliated with University of Washington and other places

What is this page?


This page lists works of an author who doesn't have a ResearchGate profile or hasn't added the works to their profile yet. It is automatically generated from public (personal) data to further our legitimate goal of comprehensive and accurate scientific recordkeeping. If you are this author and want this page removed, please let us know.

Publications (358)


Air pollution exposures in early life and brain development in children (ABC): protocol for a pregnancy cohort study
  • Article

February 2025

·

22 Reads

·

·

·

[...]

·

Introduction Air pollution is linked with poor neurodevelopment in high-income countries. Comparable data are scant for low-income countries, where exposures are higher. Longitudinal pregnancy cohort studies are optimal for individual exposure assessment during critical windows of brain development and examination of neurodevelopment. This study aims to determine the association between prenatal ambient air pollutant exposure and neurodevelopment in children aged 12, 24 and 36 months through a collaborative, capacity-enriching research partnership. Methods and analysis This observational cohort study is based in Nairobi, Kenya. Eligibility criteria are singleton pregnancy, no severe pregnancy complications and maternal age 18 to 40 years. At entry, mothers (n=400) are administered surveys to characterise air pollution exposures reflecting household features and occupational activities and provide blood (for lead analysis) and urine specimens (for polycyclic aromatic hydrocarbon (PAH) metabolites). Mothers attend up to two additional antenatal study visits, with urine collection, and infants are followed through age 36 months for annual neurodevelopment and caregiving behaviour assessment, and child urine and blood collection. Primary outcomes are child motor skills, language and cognition at 12, 24 and 36 months, and executive function at 36 months. The primary exposure is urinary PAH metabolite concentrations. Additional exposure assessment in a subset of the cohort includes residential indoor and outdoor air monitoring for fine particulate matter (PM2.5), carbon monoxide (CO), ultrafine particles (UFP) and black carbon (BC). Ethics and dissemination This study was approved by the Kenyatta National Hospital - University of Nairobi Ethics and Research Committee, and the University of Washington Human Subjects Division. Results are shared at annual workshops.


Comparison of marginal benefits of reducing PM2.5 concentrations in a hypothetical city of 250,000 people across different estimates of the C-R relationship. Marginal benefits are calculated using a VSL of $8.6 million. Left panel, Krewski et al. (2009) log-linear versus log-log. Middle panel, Krewski et al. (2009) log-linear versus Lepeule et al. (2012) log-linear. Right panel, Krewski et al. (2009) log-log versus Burnett et al. (2018) which is based on a flexible functional form. The population-weighted PM2.5 concentration distribution for the US is shown by the bars at the bottom of each panel
Marginal abatement cost and marginal benefit functions from unit 1 from ALCOA Power Plant, Warrick, IN. Step functions, in red, are marginal abatement cost functions for each pollutant with the name of each abatement option listed on each step. The blue lines show the marginal benefits of abatement from this unit, at the mean marginal benefits (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{I}_{MB}=1$$\end{document})
Share of initial damages remaining after optimal control technologies implemented for each of nine scenarios (white dots). Krewski log-linear (left) and Krewski log-log (right) functional forms for C-R. The fitted lines demonstrate a close approximation of the optimal damages for any value of the marginal benefit index. The box and whisker plot, inside the graph, shows the distribution of the marginal benefit index across draws. The box and whisker plot, outside the axes, shows the distribution of optimal damages corresponding to the marginal benefit index. The green line is the mean, the black line is the median, the box represents the interquartile range, and the whiskers extend to the 5th and 95th percentiles
Share of initial emissions under optimal control technology across nine scenarios, for SO2, NOX and PM2.5 emissions. Red and blue lines are fitted to the data points (white circles)
Change in optimal damages given different background PM2.5 concentrations for log-linear (blue) and log-log (red). The lines are fitted to the data points (white circles). The solid lines and data points represent the results when background concentrations are at baseline levels (100%). The lighter lines and data points represent the results when background concentrations are at 50% of the baseline

+3

Optimal Point Source Abatement Technology Adoption: The Impact of Uncertainty in the Benefits of Abatement
  • Article
  • Publisher preview available

January 2025

·

9 Reads

Environmental and Resource Economics

Reducing emissions from point sources may be justified by the large expected benefits of improved health. However, the optimal reduction in emissions is complicated by the large uncertainty regarding the magnitude of these benefits. In particular, there is uncertainty in the size of the impact of pollution on increased premature mortality, and in the monetary valuation of reducing risks of mortality. We calculate the optimal emission reductions from abatement technology adoption at most point sources of SO2, NOX, and primary PM2.5 in the United States across a wide range of uncertainty in the parameters used to estimate benefits of reductions. The results demonstrate that although the range of uncertainty in benefits is very wide, as long as the benefits are not at the low end of the distribution, the optimal abatement from sources is in a relatively narrow range. It is when benefits of reducing pollution are well below their mean estimates that the optimal reduction in emissions varies substantially. Resolving the likelihood of very low benefits of abatement could potentially reduce the uncertainty regarding optimal abatement policy.

View access options


Characterizing Indoor-Outdoor PM2.5 Concentrations Using Low-Cost Sensor Measurements in Residential Homes in Dhaka, Bangladesh

November 2024

·

113 Reads

Atmospheric Environment

We collected simultaneous indoor and outdoor PM2.5 measurements from 17 homes in Dhaka, Bangladesh, to characterize spatio-temporal variations and identify factors influencing indoor and outdoor PM2.5 levels. A pair of PurpleAir PM2.5 sensors were deployed at each home, one indoors and the other outdoors, during the wet and dry seasons, and the locally calibrated data were used for analysis. Indoor and outdoor PM2.5 levels were three times higher during the dry season (indoor 146 ± 22 μg/m³, outdoor 153 ± 23 μg/m³) than during the wet season (indoor 52 ± 12 μg/m³, outdoor 50 ± 11 μg/m³). Indoor to outdoor (I/O) ratios were close to 1 in both seasons (dry: 0.97 ± 0.14, wet: 1.05 ± 0.19). This suggests that regional background pollution levels significantly influence indoor levels observed in different households. Infiltration factors closer to 1 (dry: 0.83 ± 0.12; wet: 0.87 ± 0.14), determined through mixed-effects regression of indoor and outdoor time series data, further highlight the substantial impact of outdoor pollution on indoor levels. Data from individual households exhibited strong temporal correlation between indoor and outdoor levels in both seasons (Pearson R: 0.82 ± 0.12 during the dry season and 0.83 ± 0.14 during the wet season), whereas indoor-outdoor spatial correlations across measured households were moderate (R: 0.49 and 0.62 during dry and wet seasons, respectively). These spatial correlations and empirical regression modeling suggest that while the spatial variation of outdoor PM2.5 levels significantly influences indoor levels' spatial variation, other factors such as indoor source activities and ventilation-related features play crucial roles in explaining variabilities in indoor PM2.5 across homes. Overall, our study suggests that indoor environments in Dhaka city are nearly as polluted as outdoor settings, and this locally derived scientific evidence can be valuable for enhancing public awareness and developing mitigation measures to reduce PM2.5 exposures in Bangladesh.



U.S. ambient air monitoring network has inadequate coverage under new PM2.5 standard

October 2024

·

21 Reads

The Clean Air Act (CAA) in the United States relies heavily on regulatory monitoring networks, yet monitoring sites are sparsely located, especially among historically disadvantaged communities. For ambient fine particulate matter (PM2.5), we compare the air quality monitoring data with spatially complete concentrations derived from empirical models to quantify the gaps of existing U.S. monitoring networks in capturing concentration hotspots and exposure disparities. Recently, the U.S. Environmental Protection Agency adopted a more stringent annual-average air quality standard for PM2.5 (9 µg/m3). Here, we demonstrate that 44% of urban areas exceeding this new standard – encompassing ~ 20 million people – would remain undetected because of gaps in the current PM2.5 monitoring network. Crucially, we find that “uncaptured” hotspots, which contain 2.8 million people in census tracts that are misclassified as in attainment of the new PM2.5 standard, have substantially higher percentages of minority populations (i.e., people of color, disadvantaged communities, and low-income populations) compared to the overall US population. To address these gaps, we highlight 10 priority locations that could reduce the population in the uncaptured hotspots by 67%. Overall, our findings highlight the urgent need to address gaps in the existing monitoring network.


PM2.5 exposure disparities persist despite strict vehicle emissions controls in California

September 2024

·

7 Reads

·

1 Citation

Science Advances

As policymakers increasingly focus on environmental justice, a key question is whether emissions reductions aimed at addressing air quality or climate change can also ameliorate persistent air pollution exposure disparities. We examine evidence from California’s aggressive vehicle emissions control policy from 2000 to 2019. We find a 65% reduction in modeled statewide average exposure to PM 2.5 from on-road vehicles, yet for people of color and overburdened community residents, relative exposure disparities increased. Light-duty vehicle emissions are the main driver of the exposure and exposure disparity, although smaller contributions from heavy-duty vehicles especially affect some overburdened groups. Our findings suggest that a continued trend of emissions reductions will likely reduce concentrations and absolute disparity but may not reduce relative disparities without greater attention to the systemic factors leading to this disparity.


High-resolution geospatial database: national criteria-air-pollutant concentrations in the contiguous U.S., 2016 – 2020

September 2024

·

2 Reads

·

1 Citation

Concentrations estimates for ambient air pollution are used widely in fields such as environmental epidemiology, health impact assessment, urban planning, environmental equity, and sustainability. This study builds on previous efforts by developing an updated high-resolution geospatial database of population-weighted annual-average concentrations for six criteria air pollutants (PM2.5, PM10, CO, NO2, SO2, O3) across the contiguous U.S. during a five-year period (2016-2020). We developed Land Use Regression (LUR) models within a partial-least-square – universal kriging framework by incorporating several land use, geospatial, and satellite – based predictor variables. The LUR models were validated using conventional and clustered cross-validation, with the former consistently showing superior performance in capturing the variability of air quality. Most models demonstrated reliable performance (e.g., mean squared error – based R2 > 0.8, standardized root mean squared error < 0.1). We used the best modeling approach to develop estimates by Census Block, which were then population-weighted averaged at Census Block Group, Census Tract, and County geographies. Our database provides valuable insights into the dynamics of air pollution, with utility for environmental risk assessment, public health, policy, and urban planning.


PM2.5 exposure disparities persist despite strict vehicle emissions controls in California

August 2024

As policymakers increasingly focus on environmental justice, a key question is whether emissions reductions aimed at addressing air quality or climate change can also ameliorate persistent air pollution exposure disparities. We examine evidence from California’s aggressive vehicle emissions control policy from 2000-2019. We find a 65% reduction in modeled statewide average exposure to PM2.5 from on-road vehicles, yet for people of color and overburdened community residents, relative exposure disparities increased. Light-duty vehicle emissions are the main driver of the exposure and exposure disparity, although smaller contributions from heavy-duty vehicles especially impact some overburdened groups. Our findings suggest that a continued trend of emissions reductions will likely reduce concentrations and absolute disparity but may not reduce relative disparities without greater attention to the systemic factors leading to this disparity.


Figure 1: Location of measurement sites and surrounding outdoor land use features. (A) The map displays the locations of selected homes for indoor and outdoor PM 2.5 measurements. Red symbols indicate locations where measurements were collected during both dry and wet seasons (S1 to S13), while black symbols represent sites where measurements from only one season were feasible (S14 to S17). Additionally, the map indicates the positions of two continuous air monitoring stations (CAMS) within Dhaka city. (B) The distribution of outdoor land use features surrounding the measurement locations, including major road (4 lanes and above) density,
Figure 2: Measured indoor and outdoor PM2.5 concentration levels and indoor-to-outdoor (I/O) ratios in individual homes during dry and wet seasons. (A) Box-whisker plot showing the distribution of hourly average indoor and outdoor concentrations measured at each home during the dry season. Data from outdoor CAMS stations for the period of measurements collected from different homes are shown. (B) I/O ratios at each home during the dry season. (C) and (D) are similar to panel (A) and (B) respectively, showing the measurement data from the wet season. For the box-whisker plot, boxes indicate the interquartile range, whiskers represent the 5 th -95 th percentile range, horizontal lines within the boxes indicate the median, and circles represent the mean. The horizontal dashed line in panels B and D serves as a visual guide.
Figure 3: Diurnal variation of indoor and outdoor PM2.5 concentration levels and indoor-to-outdoor (I/O) ratios during dry and wet seasons. (A) Diurnal variation of indoor PM2.5 levels during the dry season. The line represents the mean diurnal profile across all sampled homes, while the shaded region indicates the range of average diurnal profiles from individual homes. (B) Similar to panel A, showing the diurnal variation of outdoor PM2.5 concentrations during the dry season. Diurnal profiles from two outdoor CAMS locations are also shown. (C) Similar to panel A, showing the diurnal variations of indoor-to-outdoor (I/O) ratios during the dry season. Panels (D), (E), and (F) are similar to (A), (B), and (C), respectively, showing measurements from wet seasons.
Summary of indoor and outdoor PM2.5 concentrations measured in individual homes during dry and wet seasons.
Multiple linear regression model for predicting spatial variabilities in indoor PM2.5 levels across sampled homes
Characterizing Indoor-Outdoor PM2.5 Concentrations Using Low-Cost Sensor Measurements in Residential Homes in Dhaka, Bangladesh

August 2024

·

55 Reads

We collected paired measurements of indoor and outdoor PM2.5 concentrations at 17 homes in Dhaka, Bangladesh, to quantify indoor-outdoor levels, their spatio-temporal variations, and influencing factors. A pair of PurpleAir PM2.5 sensors were deployed at each home, one indoors and the other outdoors, during the wet (June to August 2021) and dry (December 2021 to February 2022) seasons, and the locally calibrated (against a beta attenuation monitor) and quality-assured data were used for analysis. Indoor and outdoor PM2.5 levels were three times higher during the dry season (indoor 146 ± 22 µg/m³, outdoor 153 ± 23 µg/m³) than during the wet season (indoor 52 ± 12 µg/m³, outdoor 50 ± 11 µg/m³). Indoor to outdoor (I/O) ratios were close to 1 in both seasons (dry: 0.97 ± 0.14, wet: 1.05 ± 0.19). This suggests that regional background pollution levels significantly influence indoor levels observed in different households. Higher infiltration factors (dry: 0.83 ± 0.12; wet: 0.87 ± 0.14), determined through mixed effect regression of parallel indoor and outdoor timeseries data, further highlight the substantial impact of outdoor pollution on indoor levels. Data from individual households exhibited strong temporal correlation between indoor and outdoor levels in both seasons (Pearson R: 0.82 ± 0.12 during the dry season and 0.83 ± 0.14 during the wet season), whereas indoor-outdoor spatial correlations across measured households were moderate (R: 0.49 and 0.62 during dry and wet seasons, respectively). These spatial correlations and empirical regression modeling suggest that while the spatial variation of outdoor PM2.5 levels significantly influences indoor levels' spatial variation, other factors such as indoor source activities and ventilation-related features play crucial roles in explaining variabilities in indoor PM2.5 across homes. Overall, our study suggests that indoor environments in Dhaka city are nearly as polluted as outdoor settings, and this locally derived scientific evidence can be valuable for enhancing public awareness and developing mitigation measures to reduce PM2.5 exposures in Bangladesh.


Citations (48)


... Similarly, whereas absolute disparity in grid emissions decreases over time for all racial-ethnic groups, relative disparity remains largely unchanged, with Black populations being the most exposed. These results have been validated by a recent study, Koolik et al. 45 , which found that California's vehicle emissions reduction policy reduced overall air pollution exposure but increased relative exposure disparities for the most exposed communities and racial-ethnic groups. Camilleri et al. 33 found similar results to our health analysis with truck electrification in an urban location, Chicago, resulting in avoided premature mortality. ...

Reference:

Impact of truck electrification on air pollution disparities in the United States
PM2.5 exposure disparities persist despite strict vehicle emissions controls in California
  • Citing Article
  • September 2024

Science Advances

... At any instant across a city, contaminants are emitted, diluted, transformed, and removed ( Fig. 1). The interplay of these processes gives rise to variations along three distinct dimensions: space, time, and chemical composition (5). Emissions sources, such as traffic and industry, are often unevenly and unequally distributed in space (6). ...

Integrating Mobile and Fixed-Site Black Carbon Measurements to Bridge Spatiotemporal Gaps in Urban Air Quality

Environmental Science and Technology

... However, they often lacked repeated sample collections, limiting their effectiveness in capturing spatio-temporal dynamics (Blanco et al., 2023;Li et al., 2019;Saha et al., 2019). Systematic repeated short-term sampling across multiple seasons at 35 locations in Dhaka city, Bangladesh, revealed a moderate intra-urban spatial gradient for PM 2.5 and a large gradient for ultrafine particle number concentration (PNC) (Saha et al., 2024). ...

Contrasting intra-urban variability of ultrafine particle number and fine particle mass concentrations in Dhaka, Bangladesh, and Pittsburgh, USA
  • Citing Article
  • April 2024

Atmospheric Environment

... Following recent studies (e.g. [23][24][25]), we use the InMAP Source Receptor Matrix (ISRM) [10,26] to estimate the change in PM 2.5 concentrations due to air pollution emissions. A full description of InMAP and the ISRM is in SI section S1.7. ...

Distributional impacts of fleet-wide change in light duty transportation: mortality risks of PM2.5 emissions from electric vehicles and Tier 3 conventional vehicles

... (16)(17)(18). This approach is sometimes called a data-only approach because mapping is accomplished by data reduction of the observations, without relying on additional predictive or mechanistic models or other input datasets (18,19). By definition, this approach only produces concentration estimates along the network of sampled road segments. ...

Multi-season mobile monitoring campaign of on-road air pollution in Bengaluru, India: High-resolution mapping and estimation of quasi-emission factors
  • Citing Article
  • January 2024

The Science of The Total Environment

... 15 Schollaert et al. designed prescribed fire management scenarios with different management extents. 30 The study simulated several decades of prescribed fire decisions and considered the evolution of fuels by applying a landscape change model. 31 The fire emissions estimated by a fuel type and combustion model were used in an inert tracer dispersion model. ...

Quantifying the smoke-related public health trade-offs of forest management

Nature Sustainability

... The key contribution of this work is to assess how less polluting conventional vehicles (Tier-3 ICVs) would fare when compared to electric vehicles in terms of air pollution and distributional equity across the nation. While past studies have detailed the increasing stringency of emissions over time [21,55] and their positive impact on air quality [56][57][58], our work incorporates Tier-3 vehicles and provides important regional conclusions. ...

PM2.5 exposure disparities persist despite strict vehicle emissions controls in California
  • Citing Preprint
  • December 2023

... Addressing longstanding and persistent air pollution inequities requires advancement in data and methods (Gohlke et al. 2023) but also in policy, with a need for both source-specific (Tessum et al. 2021) and location-specific (Wang et al. 2022) strategies to eliminate inequities in exposure. ...

State-of-the-Science Data and Methods Need to Guide Place-Based Efforts to Reduce Air Pollution Inequity
  • Citing Article
  • December 2023

Environmental Health Perspectives

... The sand cat optimization algorithm (SCOA) is modified and employed for ANN's hyper-parameter optimization. Feldman et al. [28] develop DL models. This method detects objects and motion using segmentation and optical flow and utilizes a regression CNN to deduce concentrations of pollutants. ...

Urban Air-Quality Estimation Using Visual Cues and a Deep Convolutional Neural Network in Bengaluru (Bangalore), India

Environmental Science and Technology

... EPA regulatory monitors comprise the largest network of high-quality ground-based PM 2.5 data that are routinely used for a variety of purposes, including as input for chemical transport models (e.g. [23]), as the ground-based measurements for developing PM 2.5 measurements from remote sensing-derived aerosol optical depth (AOD) [24,25], to assess the accuracy [26] and calibration of low-cost sensors [27], and as the input dataset for empirical ("land-use regression") models -including for the six major models producing national estimates [28]. As Bechle et al. (2023) [28] state "whatever strengths or weaknesses exist in using EPA monitors (and their locations) for empirical models, those likely impact all of the [empirical] models." ...

Intercomparison of Six National Empirical Models for PM2.5 Air Pollution in the Contiguous US

Findings